Follow-up Comment #18, bug #66342 (group groff): At 2026-02-03T08:03:54-0500, Deri James wrote: > Follow-up Comment #17, bug #66342 (group groff): > I have had continual timeout connections to https://savannah.gnu.org > since the weekend.
Ugh. That really sucks. :( > Is it just me! :-) I'm not having that exact problem, but I've had others like it. https://savannah.nongnu.org/support/?111374 If you can't get there, you can't see it, but it's a ticket from Benno Schulenberg complaining of stale cgit.git.savannah.gnu.org mirrors. Load-balancing gone wrong, perhaps. Benno wrote: >>> Looking at https://cgit.git. ... rg/cgit/nano.git/ it shows "copyright: >>> update the years for the FSF" from two weeks ago as the last commit. >>> But yesterday it showed "docs: add an Exit Status section to the man >>> page and the manual" from yesterday as the last commit. >>> >>> When I now do `git push` on that last commit, it says: Everything >>> up-to-date. But the cgit still shows the state from two weeks ago. I added: >>> Concur. One of the cgit mirror sites was stale for groff by about >>> two whole weeks! >>> >>> I tried sniffing out the culprit by running "nslookup" on >>> "cgit.git.savannah.org" and putting the IP address directly in the >>> URLs, but this didn't work due to HTTPS default and inappropriate or >>> missing SSL certificate matches. (I'm not au fait with modern web >>> security.) >>> >>> What made this worse was that mirror selection seems to be arbited >>> on source-IP basis, so restarting my browser did no good--if my >>> browser picks the stale mirror, that is the one I see from that host >>> for an indefinitely long period (several days). >>> >>> I got multiple IP addresses from "nslookup" on >>> "cgit.git.savannah.org", as I pretty much expected. >>> >>> I'm no load-balancing expert, but that was the technique I was >>> expecting. > But from what I remember of your sensible comment #8 I was in > agreement and stuck it on my todo list after 1.24.0 is out the door. Here's Dave's comment #8 in its entirety: >>> Deri, >>> >>> Would it make sense to make embedding the default behavior, continue >>> to accept "-e" as a no-op (as Branden did), and introduce a new flag >>> to specify not embedding the base 14 fonts, for those users who do >>> need smaller PDFs? This would change existing behavior for users >>> who do not specify -e, but in a way more in line with best >>> practices, and would not affect users who do specify -e. Is that an >>> acceptable change? My amendment would be to give that "new flag" an identity of `-E`. But I would also like to see the meaning of bit 3 of `--opt` inverted. I'd like to see this done _before_ 1.24.0 is out, to avoid the necessity of changing the meaning of bit 3 of gropdf's `--opt`. Unless of course you disagree with that proposal, in which case further discussion is warranted. If you agree, I'm happy to help develop the changes. > If you have access to savannah please can you confirm this message > (via the email route to savannah) has arrived. Confirmed; I see it on the Savannah website as comment #17 to bug #66342. > I have a hankering You twigged it, rather. ;-) A hankering is a feeling of desire. "Hmm, I guess Bart's not to blame. He's lucky too, because it's spankin' season and I've got a hankerin' for some spankerin'!" > I may have triggered an anti ddos rule and got myself blacklisted, I > believe the last time I successfully accessed the site at the weekend, > firefox was mis-behaving, I left it to have my lunch and when I came > back it was still using a lot of cpu and I killed it, but I wonder if > during that high cpu it was hammering POST messages at savannah, which > put me on "the naughty step". Maybe, but I've also fallen victim to (what I suspect to be) anti-DDoS measures simply by clicking through GNU mailing list archives "too fast". "Too fast" is, I guess, on the order of under 1 second per HTTP(S) request. > I have recently (my xmas present to myself) upgraded my internet > connection to 500Mbs, so firefox misbehaving would be serious. Yes. A disadvantage of a high data rate as a consumer is that we become more confusable with deep-pocketed crawler bots who are trying to masquerade as ordinary consumers. > Do you have an email address I can use to contact savannah admins who > can help me get access again. Of course, it could be savannah is under > a real ddos attack and I've just been unlucky not to get a connection > slot. Yes, you can contact (and I recommend subscribing to) <[email protected]>. There's also this resource: https://hostux.social/@fsfstatus ...but in my experience neither of these is updated as frequently as outages occur in practice. I've more than once had to mail that -public list to solicit information. On the upside, that usually works, but I don't usually get a response super-fast. _______________________________________________________ Reply to this item at: <https://savannah.gnu.org/bugs/?66342> _______________________________________________ Message sent via Savannah https://savannah.gnu.org/
signature.asc
Description: PGP signature
